Differential sensitivity of Ca 2+ wave and Ca 2+ spark events to ruthenium red in isolated permeabilised rabbit cardiomyocytes

Transcription

1 J Physiol (2010) pp Differential sensitivity of Ca 2+ wave and Ca 2+ spark events to ruthenium red in isolated permeabilised rabbit cardiomyocytes N. MacQuaide, H. R. Ramay, E. A. Sobie and G. L. Smith Institute of Biomedical and Life Sciences, West Medical Building, University of Glasgow, Glasgow G12 8QQ, UK Spontaneous Ca 2+ waves in cardiac muscle cells are thought to arise from the sequential firing of local Ca 2+ sparks via a fire diffuse fire mechanism. This study compares the ability of the ryanodine receptor (RyR) blocker ruthenium red (RuR) to inhibit these two types of Ca 2+ release in permeabilised rabbit ventricular cardiomyocytes. Perfusing with 600 nm Ca 2+ (50 μm EGTA) caused regular spontaneous Ca 2+ waves that were imaged with the fluorescence from Fluo-5F using a laser-scanning confocal microscope. Addition of 4 μm RuR caused complete inhibition of Ca 2+ waves in 50% of cardiomyocytes by 2 min and in 100% by 4 min. Separate experiments used 350 μm EGTA (600 nm Ca 2+ ) to limit Ca 2+ diffusion but allow the underlying Ca 2+ sparks to be imaged. The time course of RuR-induced inhibition did not match that of waves. After 2 min of RuR, none of the characteristics of the Ca 2+ sparks were altered, and after 4 min Ca 2+ spark frequency was reduced 40%; no sparks could be detected after 10 min. Measurements of Ca 2+ within the SR lumen using Fluo-5N showed an increase in intra-sr Ca 2+ during the initial 2 4 min of perfusion with RuR in both wave and spark conditions. Computational modelling suggests that the sensitivity of Ca 2+ wavestorurblockdependsonthenumberofryrsper cluster. Therefore inhibition of Ca 2+ waves without affecting Ca 2+ sparks may be explained by block of small, non-spark producing clusters of RyRs that are important to the process of Ca 2+ wave propagation. (Resubmitted 20 May 2010; accepted after revision 30 September 2010; first published online 4 October 2010) Corresponding author G. L. Smith: Institute of Biomedical and Life Sciences, West Medical Building, University of Glasgow, Glasgow G12 8QQ, UK. g.smith@bio.gla.ac.uk Abbreviations reticulum. RuR, ruthenium red; SERCA, sarcoplasmic/endoplasmic reticulum Ca 2+ -ATPase; SR, sarcoplasmic Introduction In ventricular cardiomyocytes, a small influx of Ca 2+ via thel-typeca 2+ channel triggers a larger global release of Ca 2+ from the sarcoplasmic reticulum (SR) inducing contraction. Release from the SR is mediated by Ca 2+ release channels (ryanodine receptor type 2, RyR) that exist in clusters of channels 2 μm apart along the longitudinal axis of the cell and 1 μm apart radially (Chen-Izu et al. 2006). The number of RyRs in each cluster is uncertain, with estimates ranging from 10 to 270 (Bridge et al. 1999; Franzini-Armstrong et al. 1998, 1999; Soeller et al. 2007). The RyRs in a cluster are thought to be functionally coupled such that the status of one channel (e.g. open) enhances the rate constant for opening of neighbouring RyR channels in the cluster (Marx et al. 2001; Yin et al. 2005). This cooperative activity is thought to be important in determining the duration of the release event (Sobie et al. 2002) andmaybe alteredby phosphorylation of RyR either by A-kinase or CaM-kinase (Marx et al. 2001; Yin et al. 2005). Ca 2+ release from individual clusters can be observed experimentally as Ca 2+ sparks (Cheng et al. 1993). These localised spontaneous Ca 2+ release events are thought not only to underlie the basic Ca 2+ release unit of excitation contraction (E C) coupling, but also to mediate a significant fraction of diastolic SR Ca 2+ release (Santiago et al. 2010). Under conditions of cellular Ca 2+ overload, larger spontaneous Ca 2+ release events known as Ca 2+ waves are observed. The Ca 2+ wave is thought to be initiated when a spontaneous Ca 2+ spark triggers the regenerative propagation of Ca 2+ release from one RyR cluster to another. Release eventually propagates through the entire heart cell via a saltatory, fire diffuse fire mechanism (Keizer & Smith, 1998). This regenerative form of SR Ca 2+ release during diastole can depolarise DOI: /jphysiol

2 4732 N. MacQuaide and others J Physiol the sarcolemma via Ca 2+ -sensitive currents, and has been implicated in initiation of ventricular arrhythmias (Weir & Hess, 1984). Recent work suggests that an alternative mode of Ca 2+ leak may occur via smaller clusters of RyRs distributed between the Z-lines outside the dyadic junction (Lukyanenko et al. 2007). It is hypothesized that spontaneous openings of these non-junctional or rogue RyRs do not produce detectable Ca 2+ sparks but contribute to RyRs Ca 2+ leak and may influence Ca 2+ wave propagation. Computational modelling of the activity of smaller clusters of RyRs that may be appropriate for rogue RyRs suggests that the reduced cooperative activity in these clusters would result in a higher sensitivity to manipulations that affect RyR activity, such as phosphorylation (Sobie et al. 2006). In this study, we investigated the effects of an RyR inhibitor, ruthenium red (RuR), on Ca 2+ sparks and waves. We present data which demonstrate that Ca 2+ wave activity is more sensitive than Ca 2+ spark activity to inhibition by RuR. The more rapid inhibition by RuR is not due to a more rapid depletion of the SR when Ca 2+ waves occur. Modelling the effect of blocking individual channels within RyR clusters of variable size predicts that those containing high numbers (e.g. 30) of RyRs resist the effects of accumulating block better than those with fewer RyRs (e.g. 5). These data suggest that Ca 2+ waves require the participation of small (sub-spark) clusters of RyR for robust initiation and propagation. Methods Cell isolation and permeabilisation All procedures and were approved by the local (University of Glasgow) ethical committee complied with both Home Office regulations and the policies of The Journal of Physiology as set out by (Drummond, 2009). New Zealand White rabbits (2 2.5 kg) were given an intravenous injection of 500 U heparin together with an overdose of sodium pentobarbitone (100 mg kg 1 ). The hearts were rapidly excised, weighed and cannulated onto a Langendorff perfusion column via the aorta. Ventricular myocytes were isolated from Langendorff perfused rabbit hearts by enzymatic digestion as previously described (McIntosh et al. 2000) and kept in a modified Krebs solution buffered with no Ca 2+ added at a concentration of 10 4 cells ml 1 until use. The cells were allowed to settle onto the coverslip at the base of a small bath. β-escin (Sigma, St Louis, MO, USA) was added from a freshly prepared stock solution to the cell suspension to give a final concentration of 0.1 mg ml 1 for min and the β-escin subsequently removed by perfusion with a mock intracellular solution with the following composition (mm): 100 KCl, 5 K 2 ATP,5Na 2 CrP, 5.5 MgCl 2,25Hepes,0.05K 2 EGTA, ph 7.0 (20 21 C). [Ca 2+ ] measurements A mock intracellular solution containing 10 μm Fluo-5F, 600 nm free Ca 2+ and either 50 μm EGTA (to allow Ca 2+ -waves) or 350 μm EGTA (to suppress waves but allow Ca 2+ sparks). Intracellular fluorescence was monitored using a Bio-Rad 2000 laser scanning confocal microscope (LSCM). The fluorophore was excited at 488 nm (Ar laser) and measured at >515 nm using linescanmodeatarateof500hz.thecellswereimaged using the epifluorescence optics of a Nikon Eclipse inverted microscope with a 60 water objective lens (NA 1.2). The iris diameter was set at 1.9, providing an axial (z) resolution of about 0.9 μm andx y resolution of about 0.5 μm based on full-width, half-maximal amplitude measurements of images of 0.1 μm fluorescent beads (Molecular Probes/Invitrogen). Data were acquired in line-scan mode at 2 ms per line scan; the pixel dimension was 0.3 μm (512 pixels per scan; zoom 1.4). The scanning laser line was oriented parallel to the long axis of the cell and placed approximately equidistant between the outer edge of the cell and the nucleus/nuclei, to ensure the nuclear area was not included in the scan line. To enable this trace to be converted to free Ca 2+ concentration ([Ca 2+ ]) a series of calibration solutions were used at the end of each experiment incorporating 10 mm EGTA as previously described (Currie et al. 2004). Ca 2+ sparksweremeasuredinperiodsof15severyminute. Ca 2+ spark parameters were measured using MacSpark: ( academicstaff/godfreysmith/macspark/), a program that used a previously established algorithm for spark detection (Cheng et al. 1999) and the detection criterion (CRI value) was set at 3.5 after an initial 5 median filter. Spark frequency was calculated per 100 μms 1. The influence of false event detection was minimised by subtracting the number of events detected in an identical perfusing solution after the cell had been exposed to a 1.5 min application of caffeine (10 mm) and thapsigargin (25 μm). Normally, Ca 2+ sparks are measured at a cytoplasmic [Ca 2+ ] of nm Ca 2+, and under these conditions sparks can easily be detected because of the low background fluorescence (using Fluo-3/4). But for this study it was important to measure Ca 2+ sparks under similar conditions to those used to examine Ca 2+ waves. Therefore we increased the Ca 2+ buffer (EGTA) sufficiently to prevent Ca 2+ waves but maintained the perfusing [Ca 2+ ] at 600 nm. Spark detection under these conditions is technically difficult due to high background fluorescence. Using Fluo-3 as an indicator the background fluorescence at 600 nm would be 3 higher than that at 120 nm and the noise on the fluorescence signal would increase by 2. In this study Fluo-5F (K d = 1.1 μm) (Loughreyet al. 2002) was used rather than Fluo-3 (K d = 0.55 μm) toreducethe

3 J Physiol Ca 2+ wave and Ca 2+ spark events 4733 background fluorescence signal and the associated noise at 600 nm thereby facilitating spark detection. The use of Fluo-5F as opposed to Fluo-3 reduced the background fluorescence and associated noise at 600 nm by 30%. Despite the lower background signal the events were difficult to detect by eye, and therefore detected events including the perimeter in linescan are highlighted using a white fill as indicated in Supplementary Fig. 1. This format is used to display linescan images containing detected spark events (Fig. 3). In experiments to monitor SR Ca 2+, cells were incubated with Fluo-5N-AM (10 μm) for 3 h at 37 C. Changes in SR Ca 2+ are expressed as relative change in caffeine-sensitive fluorescence (F/F o,cs ), where the data are normalised to the inter-wave fluorescence level after subtraction of the level measured during caffeine application (F o,cs ). Fluorescence signals were monitored using linescan parameters used to measure a cytoplasmic signals to ensure that the Fluo-5N signal tracked the Ca 2+ wave. Long term measurements of luminal Ca 2+ signal were made using the confocal microscope in frame-scan mode every 30 s with the pin-hole aperture opened to the maximum. This procedure gave good signal/noise with minimal bleaching over a min of continuous data capture. Ca 2+ wave characteristics were obtained using semi-automated analysis software written by one of the authors (N.M.). Modelling Mathematical modelling was used to gain insight into how inhibition of RyRs by RuR affected Ca 2+ spark properties and Ca 2+ wave propagation. The primary goal was to determine what characteristics of RyR clusters were required to reproduce the unexpected observation that RuR abolished Ca 2+ waves while simultaneously causing only minor changes in Ca 2+ spark frequency and amplitude (see below). To avoid conclusions that depended on specific RyR gating mechanisms, we purposely employed a simplified model with a limited number of free parameters. The basic assumption of the modelling was that the drug renders a certain percentage of channels within the RyR cluster inactive. Thus, the behaviour of a RyR cluster with (for example) 20% of the channels inhibited by RuRisequivalenttothatofaclusterwith20%fewer channels physically present. Given this assumption, we systematically varied the number of channels in the RyR cluster to generate predictions of how RuR affected spark amplitude and Ca 2+ wave propagation. ThefirststepwastopredictthelocalCa 2+ release fluxes resulting from RyR clusters of different sizes. The basic structure of the model used for these simulations was similar to that described previously (Sobie et al. 2002); however, to avoid complications caused by stochastic RyR gating, we assumed that all RyRs opened for a fixed duration during Ca 2+ sparks (as in Sobie et al. 2005). Besides the number of RyRs in the cluster, the most important free parameter in these simulations was the rate of refilling from network to junctional SR, which determined the degree of local SR depletion during the spark. Next, the calculated local Ca 2+ release fluxes were used as the input to an established model of Ca 2+ diffusion in the cytosol, binding to intracellular buffers including the fluorescent indicator Fluo-5F, and blurring by the optical apparatus (Smith et al. 1998; Sobie et al. 2002). The resulting simulated Ca 2+ sparks were analysed to derive the predicted relationship between the fraction of inactive RyRs and Ca 2+ spark amplitude. The final step was to predict how RuR-induced changes in the local SR Ca 2+ release flux affected the propagation of Ca 2+ waves. We assumed that 50 clusters were arranged linearly, and that waves failed to propagate if a single cluster out of 50 was not triggered by its immediate predecessor. We further assumed that each cluster would be activated if the total quantity of Ca 2+ released by the preceding cluster was above a certain threshold. Although this threshold was somewhat arbitrary, its absolute level could be varied over a sizeable range without affecting the main conclusions (see supplementary Fig. S1). To calculate the probability of propagation failure, we used the binomial distribution to compute the probability that at least one cluster out of 50 released less Ca 2+ than the threshold amount. For instance, if N channels are present in each cluster, and p per cent of channels are inhibited by a drug, then the probability P that exactly k channels are inhibited is: P = N! k!(n k)! p k (1 p ) N k After calculating P for all values of k, wecomputedthe probability that the inhibition of one or more clusters out of 50 was sufficient to decrease the quantity of Ca 2+ released below the threshold level. Statistics All data are presented as means ± standard error of the mean (S.E.M.) with the number of myocytes examined as n. All the results presented were statistically significant at P < 0.05 using Student s t test or one-way ANOVA. Results The effects of RuR on wave propagation Rabbit ventricular myocytes were permeabilised with β-escin and perfused with a mock intracellular solution containing a free [Ca 2+ ] of approximately 600 nm. This level of Ca 2+ induced spontaneous Ca 2+ waves with a frequency of 0.4Hzforperiodsinexcessof15min(data

4 4734 N. MacQuaide and others J Physiol not shown). This level was chosen to produce regular Ca 2+ waves every 2 3 s to allow accurate resolution of the time taken for RuR to inhibit Ca 2+ waves. Lower perfusing [Ca 2+ ] would reduce the wave frequency and thereby reduce time resolution. Higher perfusing [Ca 2+ ] would produce higher frequency waves and better time resolution but a higher background fluorescence. As shown in Fig. 1, addition of 4 μm RuR to the perfusate caused inhibition of Ca 2+ waves within 4 min. Figure 1A demonstrates the mode of wave inhibition routinely observed on application of 4 μm RuR. Initially, regular Ca 2+ waves were observed (Fig. 1A), with no apparent change in frequency or amplitude of the events. After 4 min, propagation was interrupted, after which no further globally propagating waves were observed. The increase in minimum Ca 2+ that occurs after Ca 2+ wave inhibition is due to the absence of an uptake phase that precedes the Ca 2+ -release phase of the wave. This transient period of rapid sarco/endoplasmic reticulum Ca 2+ -ATPase (SERCA)-mediated Ca 2+ uptake is sufficient to briefly reduce the [Ca 2+ ] within the permeabilised preparation below the ambient [Ca 2+ ]. When a higher concentration of RuR (15 μm) was used, propagating Ca 2+ waves were abolished in all cells within 1 min. Figure 1B summarises this relationship, showing that by 2 min of continuous perfusion with 4 μm RuR, Ca 2+ waves were blocked in 50% of cells, by 3 min waves were blocked in 85% of cells, and after 4 min propagation was blocked in 100% of cells. Prior to the moment of RuR-induced inhibition of the Ca 2+ waves, the velocity and amplitude of these events were similar to control values. Comparing the velocity and amplitude of the last three Ca 2+ waves prior to RuR addition with the last three waves prior to RuR-induced wave block, no significant differences were evident (velocity: ± 4.19 vs ± μms 1 and amplitude ( F/F o ): 3.28 ± 0.19 vs ± 0.31). Therefore there was no evidence of progressive effects before RuR-induced block. By applying caffeine (10 mm) rapidly via a glass pipette at the end of the cell, distal to the perfusion inflow, wave propagation was successfully reinstated temporarily (Fig. 2A). The local SR Ca 2+ release caused by caffeine triggered a propagating Ca 2+ wave in the remainder of the cell. The first wave initiated in this way propagated with a higher velocity (148.0 ± 16.3% of control, n = 5, P < 0.05) but was of similar amplitude (115 ± 26%). This observation, seen in all cells studied, indicated that RuR rapidly blocked the initiation of Ca 2+ waves, prior to preventing longitudinal propagation. To determine whether RuR abolished Ca 2+ waves by reducing SR [Ca 2+ ], cells were loaded with the low affinity Figure 1. Inhibition of Ca 2+ waves by RuR The time course of RuR effects on wave propagation in permeabilised myocyte. Aa, linescan showing propagating Ca 2+ waves during RuR application (upper). Red and blue F/F o signals derived from sections indicated by the red and blue bars at either side of the image above (lower). RuR blockade of RyR causes propagation to fail, followed by total failure of initiation event. Ac and d, linescans and associate F/F o from grey bars indicated in regions i and ii of Aa. B, time course of wave inhibition by RuR.

5 J Physiol Ca 2+ wave and Ca 2+ spark events 4735 indicator Fluo-5N and perfused after permeabilisation with the same solutions as used in the measurements shown in Fig. 2A and B. On addition of RuR, the averaged Fluo-5N signal increased over the initial 2 4 min (Fig. 2C). At the end of 4 min, when Ca 2+ waves had been inhibited in 100% of cells (Fig. 1B) SR[Ca 2+ ] was 116 ± 2% of control (n = 8, P < 0.05). The effects of RuR on Ca 2+ sparks Figure 3A shows that for a considerable time after cessation of Ca 2+ waves, Ca 2+ sparks still occurred at a frequency comparable to that seen in control cells ( 7.2 sparks (100 μm 1 s 1 ). In separate experiments, permeabilised cells were exposed to the same free [Ca 2+ ]( 600 nm), but with propagated release prevented by increasing cytoplasmic Ca 2+ buffering (350 μm EGTA). Figure 3B shows representative traces before and at various times following RuR application, illustrating that the number of events progressively decreased during perfusion with RuR. Figure 3Ca and b shows average profiles of sparks recorded at different time points indicating no major changes in amplitude and width despite a decrease in frequency. Figure 3Da shows mean data from all cells studied (n = 18 cells from 4 hearts). Under control conditions in the absence of RuR Ca 2+ spark frequency was 9.1 ± 1.2 sparks (100 μm) 1 s 1, and this value was obtained after subtracting the false events detected after caffeine/thapsigargin treatment (1.2 ± 0.33 sparks (100 μm) 1 s 1 ). After 2 min, Ca 2+ sparks occurred with a similar frequency to control (87.1 ± 7.0%). By 4 min of RuR perfusion, spark frequency had fallen significantly (to 39.8 ± 9.5%, P < 0.05) and finally at 10 min, spark activity was not significantly different from zero (1.4 ± 6.8% of the controlrate).figure3dc shows that no significant changes Figure 2. Caffeine responses after RuR inhibition Local caffeine application allows reinitiation of propagating Ca 2+ waves. A, linescan and F/F o showing propagation block, then local caffeine application (black arrows) reinitiates the propagated release. Ba, amore detailed view of region i from A, showing the first 2 propagating waves initiated by caffeine. Bb, region ii from A showing failure of caffeine-initiated propagation events, but still allowing SR Ca 2+ release. C, Ca 2+ measurement with SR loaded Fluo-5N-AM during Ca 2+ waves. a, SR[Ca 2+ ] rise observed expressed as F/F o,cs followed by caffeine induced Ca 2+ depletion. RuR applied 10s after beginning of trace. b, mean F/F o,cs from 8 cells.

6 4736 N. MacQuaide and others J Physiol Figure 3. Inhibition of Ca 2+ sparks by RuR Spark-mediated Ca 2+ release persists subsequent to RuR inhibition of Ca 2+ waves. A, linescan image showing cessation of Ca 2+ waves (a), with persistent spark release shown in b (see zoomed region outlined within the black rectangle). The detected events are shown in white. B, example linescan images taken at specified periods after application of RuR. Ca 2+ sparks measured under 600 nm free Ca 2+ and high (350 μm) EGTA. Images shown have been filtered using a 15 point Savitzky Golay filter (spatial) and 7 boxcar (temporal); the detected events are shown in white. C, Average temporal (a) and spatial(b) profiles of the largest 10% spark at different times after addition of RuR Da, summary plot of spark frequency vs. time normalised to each cell s control and by subtracting the spark frequency of each cell after thapsigargin (25 μm) and caffeine (10 mm) application for 2 min. Db, SR content changes (expressed as F/F o,cs ). RuR added 10 s after time zero. Dc, mean spark amplitude (upper), width (middle) and duration (lower), all plots normalised to control.

7 J Physiol Ca 2+ wave and Ca 2+ spark events 4737 in spark amplitude or duration, and only a small change in spark width, were observed as the number of detectable sparks decreased. This is in contrast to earlier work measuring Ca 2+ sparks at 100 nm where spark amplitude decreased substantially in 5 μm RuR (Lukyanenko et al. 2000). As with the Ca 2+ wave protocol, cells were loaded with Fluo-5N to assess SR [Ca 2+ ] during RuR exposure in a parallel set of measurements using identical solutions. The mean (n = 8 cells) is shown in Fig. 3Db.Fluorescence increased gradually to ± 4.0% (P < 0.05 n = 8) over a 5 min period, indicating an increase in SR content similar to that observed during the Ca 2+ wave protocol. To check that the SR load was comparable under these two conditions (600 nm Ca 2+ in the presence of 50 or 350 μm EGTA), caffeine induce Ca 2+ release was measured in the absence of RuR. Using the cytoplasmic Fluo-5F signal and allowing for the extra cytoplasmic buffering due to EGTA (MacQuaide et al. 2009), SR content was 192 ± 17 μm (l cell volume) 1 (n = 7) in the presence of 50 μm EGTA and 205 ± 32 μm (l cell volume) 1 (n = 5) in the presence of 350 μm EGTA. This indicates that there were no major differences in SR Ca 2+ content in the two experimental groups. Comparison of Ca 2+ wave and Ca 2+ spark inhibition To illustrate the differences in time course of the inhibitory action of RuR on Ca 2+ waves and Ca 2+ sparks, the relative changes in wave incidence and Ca 2+ spark frequency for each 1 min interval are plotted in Fig. 4. The data indicated that, prior to changes in Ca 2+ spark frequency, RuR was able to significantly reduce the incidence of Ca 2+ waves. This suggests that the mode of action of the RuR-induced block of Ca 2+ waves was not simply via inhibition of Ca 2+ sparks. In parallel studies, the relationship between Ca 2+ sparks and Ca 2+ waves was determined using other methods of blocking Ca 2+ waves: (i) reduced RyR sensitivity using tetracaine and (ii) SERCA inhibition via tetrabutylquinone (TBQ). Each of these interventions produced roughly parallel changes in Ca 2+ waves and Ca 2+ sparks, in marked contrast to the highly non-linear relationship observed with RuR. amplitude, a quantity closely related to the total Ca 2+ released during the spark (see supplementary Fig. S3), versus the percentage of available RyRs, assuming that either 30 RyRs or five RyRs were present in the control cluster. This illustrates the relative insensitivity to RyR availability in the former case (30 RyRs) compared to the latter (5 RyRs), which occurs because the local SR Ca 2+ depletion during sparks is faster and more complete when greater numbers of RyRs are present in the cluster. Additional calculations were performed to predict the effects of RuR block on the propagation of Ca 2+ waves. Assuming that a 35% decrease in SR Ca 2+ release in one RyR cluster prevented wave propagation (see Methods), Fig. 5C shows that when 30 RyRs per cluster is assumed, Ca 2+ wave propagation remains close to control values until 60% of RyRs are blocked by RuR. In contrast, Ca 2+ wave probability drops to zero with only moderate levels of RyR inhibition when each RyR cluster is assumed to contain only five channels. This demonstrates how cluster size can dramatically influence the predicted response to RyR blockade. Discussion In this study, the RyR blocker RuR was used to compare and contrast the time course of inhibition of spontaneous Ca 2+ waves with that of spontaneous Ca 2+ sparks. The primary mode of block of RyR by RuR is a reduction in the overall availability of channels by inducing long closures (Rousseau & Meissner, 1989; Ashley & Williams, 1990; Lindsay & Williams, 1991; Xu et al. 1998; Lukyanenko et al. 2000). Thus on addition of RuR there are fewer RyRs available for activation within each cluster at any Computational modelling of Ca 2+ wave propagation Computational modelling was used to explore potential mechanisms underlying the differential effects of RuR on Ca 2+ sparks and waves. The simulations generated predictions of how altering the number of RyRs in each cluster affected Ca 2+ spark amplitude and the probability of wave propagation. Given the assumption that RuR inhibition makes channels essentially inactive, altering the cluster size is equivalent to simulating different degrees of RuR inhibition. Figure 5B shows the predicted spark Figure 4. Relationship between wave occurrence and spark frequency measured after inhibition of RyR or SERCA Parallel experiments with 600 nm free Ca 2+, for spark and wave experiments, with respective low and high EGTA concentrations (see Methods). Experiments with inclusion of TBQ (open symbols), tetracaine (filled grey symbols) and RuR (filled black symbols) are plotted as spark frequency vs. no. of cells exhibiting waves. In both cases the measurements represent steady state effects of both TBQ and tetracaine.

8 4738 N. MacQuaide and others J Physiol time. Perfusion with RuR leads to the rapid inhibition of Ca 2+ waves despite the continued presence of Ca 2+ sparks. When the incidence of Ca 2+ waves was reduced to 50% of control values, Ca 2+ spark characteristics were not significantly altered. When wave incidence was only 5% of control, spark frequency was decreased by only 40% (Ca 2+ spark amplitude, duration and width were unchanged). These differences occurred despite similar levels of intra-sr [Ca 2+ ] under the two conditions. This highlights a previously unreported discrepancy in the response of unitary and propagating events to this mode of blockade. The Ca 2+ wave is a consequence of a series of fire diffuse fire events between clusters of RyRs after an initiating Ca 2+ spark. Therefore the data indicate that the inhibition of Ca 2+ waves by RuR is not caused by the absence of an initiating Ca 2+ spark but rather interference with the diffuse or subsequent triggering of a fire event. A possible explanation for the disparity between wave and spark activity comes from computational modelling of the behaviour of variable sizes of RyR clusters. RuR block of Ca 2+ waves cannot be explained by the action of RuR on large clusters of RyRs (>30) that generate detectable Ca 2+ sparks. Instead the sensitivity suggests the involvement of much smaller clusters of RyRs (<5 RyRs) than those thought to generate the archetypical Ca 2+ sparks (Bridge et al. 1999; Franzini-Armstrong et al. 1998, 1999; Soeller et al. 2007). Thus non-spark RyR activity may be an essential component in the formation and propagation of Ca 2+ waves. Figure 5. Probabalistic model of propagating Ca 2+ release A, illustration of the computational model of the propagation of a Ca 2+ wave along arrays of 50 RyR clusters (R1 50), containing 30 (upper) or 5 (lower) RyRs per cluster. The model examined the effects of progressive block of RyRs; when the magnitude of release from one site fell below the threshold level, propagation failed. B, data from previously published model of Ca 2+ release during a Ca 2+ spark (Sobie et al. 2002); this formulation was used to describe the Ca 2+ release at each site in the cluster. The graph shows model output of Ca 2+ spark amplitude from clusters with 5 (blue) and 30 (red) RyRs. Arbitrary threshold set for proportion of cluster RyRs available Ca 2+ activation required for propagation (indicated by the dotted line). C, model output showing percentage of cells with propagating waves, given a cluster size of 5 (blue) and 30 (red). Experimental data from Fig. 4 is superimposed for comparison. Inhibition of Ca 2+ sparks by RuR The experimental protocols used in this study made use of the slow diffusion of RuR into myocytes, and consequently RyR activity became attenuated gradually over 5 10 min, consistent with a previous study (Lukyanenko et al. 2000). This gradual time course of block allowed fine resolution of itseffectsonca 2+ sparks and Ca 2+ waves. Spark frequency was reduced to 50% within 5 min of RuR exposure, falling to 1% of control after 10 min in the presence of 600 nm Ca 2+ (350 μm EGTA). Ca 2+ spark duration and width decreased with an even slower time course, while amplitude remained relatively constant. The changes are consistent with those observed in a previous study in the presence of 150 nm Ca 2+ (100 μm EGTA) (Lukyanenko et al. 2000), with the exception of amplitude which showed a reduction to approximately one-third of control in this earlier study. Differences in bathing [Ca 2+ ] and indicator (Fluo-3 vs. Fluo-5F) may be responsible for the disparity between the two studies. In the present study, the consequence of the slow onset of effects of 4 μm RuR is that after 2 min of exposure to RuR, no significant changes in spark frequency, amplitude, width or duration were observed (i.e. changes were < 3%). Parallel measurements of SR Ca 2+ indicated that during this time SR Ca 2+ content increased. The SR Ca 2+ level is a balance between uptake via SERCA and release via Ca 2+ leak pathways, and thus a rise of SR Ca 2+ (in the absence ofsercastimulation)suggestsareductionintheflux through the Ca 2+ leak pathways, yet no changes in Ca 2+ sparks were evident. Two possible options to explain the lack of detectable changes in Ca 2+ spark characteristics are:

9 J Physiol Ca 2+ wave and Ca 2+ spark events 4739 (1) Autoregulation. Inhibition of Ca 2+ sparks would quickly lead to an increase in SR Ca 2+ which would in turn lead to an increase in RyR activity and spark frequency back to normal via SR luminal regulation of RyR activity. In the steady state, SR Ca 2+ efflux would return to normal in the presence of a higher SR Ca 2+. This type of autoregulation has been previously used to explain the biphasic action of the drug tetracaine (Gyorke et al. 1997). On addition of tetracaine, spark frequency decreased for 1 2 min before returning to the control rate over the subsequent 5 min in parallel with a rise in SR Ca 2+ levels. Given this time course, it seems unlikely that autoregulation of Ca 2+ sparks occurs in the current study since even over a period of 1 min, no detectable changes in Ca 2+ spark activity were observed, and therefore this was not the mechanism underlying increased SR Ca 2+ levels. (2) Sub-sparks. The sparks detected may represent the largest of a range of spark size. RuR might inhibit a subpopulation of smaller sparks/release events that could not be detected above the background fluorescence signal but contribute to SR leak. If these undetectable sub-sparks are crucial to Ca 2+ wave formation, their inhibition by RuR may explain the poor correlation between Ca 2+ wave events and detectable Ca 2+ sparks. In a recent study (Zima & Blatter, 2009), SR Ca 2+ efflux was proposed to occur via three distinct pathways: (i) Ca 2+ sparks via RyR clusters, (ii) a non-spark pathway mediated via RyR, and (iii) a non-spark pathway not mediated via RyR. The exact molecular identities for (ii) and (iii) are not certain, but the increase in SR Ca 2+ content in the absence of changes in Ca 2+ spark characteristics observed in the current study could be explained in terms of inhibition of a non-spark, RyR-mediated pathway. Interruption of Ca 2+ wave propagation by RuR Superfusion with RuR caused a rapid inhibition of Ca 2+ waves without obvious changes in wave velocity, amplitude or frequency. This contrasts with other manipulations known to inhibit Ca 2+ waves (Gyorke et al. 1997; Trafford et al. 2000; Smith & O Neill, 2001; MacQuaide et al. 2007). In these cases, tetracaine caused an initial slowing of wave propagation before complete block. In the continued presence of tetracaine, Ca 2+ waves returned, in parallel with ariseinsrca 2+ content (Gyorke et al. 1997). In contrast, the immediate and sustained inhibitory effect of RuR without inhibition of Ca 2+ sparks suggests interference with the events initiating a Ca 2+ wave, i.e. the diffuse or fire events associated with propagation of a Ca 2+ wave. Rapid local application of caffeine after complete inhibition of Ca 2+ waves nonetheless caused a large release of Ca 2+ from the SR. This provides further evidence that the mechanism of Ca 2+ wave inhibition by RuR is more complicated than the drug simply making all RyRs inactive. Initially, the local release was able to propagate along the length of the cell at wave velocities comparable to the Ca 2+ waves prior to RuR addition, but within 5 10 s this propagated release was inhibited. Why RuR should block wave initiation before propagation is unclear; one possible explanation is provided by a modelling study that suggests that wave initiation may involve regions of the cell with marginally shorter sarcomere lengths, where initial longitudinal propagation can occur with a higher probability (Izu et al. 2006). RuR may be able to reduce the probability of propagation in these regions before propagation between clusters in regions of longer sarcomere length. Further work is required to investigate this issue. Slowly propagating waves were not observed either just prior to block or during caffeine triggered wave production. One possible explanation is the concomitant increase of intra-sr Ca 2+ during RuR perfusion. This will result in larger amplitude Ca 2+ waves that would tend to propagate faster if no other changes were taking place. Thus the absence of significant changes in velocity may result from two opposing effects, RuR-induced inhibition of RyR activity that slows the last one to two wave events and increased SR load that tends to produce larger Ca 2+ release events. Unfortunately, the poor sensitivity of Fluo-5F signals to the [Ca 2+ ] reached at the peak of the wave prevents significant changes in Ca 2+ wave amplitude being easily detected. RuRisalsoaneffectiveinhibitorofthemitochondrial uniporter, and therefore parallel inhibition of mitochondrial Ca 2+ uptake must occur on addition of RuR. This action does not underlie the inhibition of Ca 2+ waves in the present study because: (i) similar effects of RuR were observed when Ca 2+ waves were initiated in the presence of mitochondrial inhibitors (1 μm FCCP, 1 μm oligomycin, data not shown) and (ii) rapid application of 10 μm Ru360, a drug with similar effects on the mitochondrial uniporter but no effects on RyR, did not block Ca 2+ waves (data not shown). Differential sensitivity of sparks and waves When the frequency of Ca 2+ sparks at particular times after addition of RuR was plotted against the incidence of Ca 2+ waves at the same time points, a highly non-linear relationship was observed (Fig. 4). This was in marked contrast to the relationships observed in the steady state (8 10 min) with increasing concentrations of the RyR inhibitor tetracaine or the SERCA inhibitor TBQ. In either the former case, with progressive reduction in RyR sensitivity, or the latter, with progressive SR Ca 2+ pump activity, the inhibition of Ca 2+ waves paralleled a decrease in Ca 2+ spark frequency. In particular, in the presence of

10 4740 N. MacQuaide and others J Physiol tetracaine, Ca 2+ waves remained despite a considerable decrease in Ca 2+ spark frequency. This seems in direct contrast to the effects of RuR, but the action of the two inhibitors on RyR are quite different. Previous work has shown that tetracaine reduces the open probability of RyR, but this effect can be reversed by an increase in SR Ca 2+ (Gyorke et al. 1997). Therefore in the steady state, RyR activity was restored, albeit at a higher SR [Ca 2+ ]. No such competitive inhibition has been shown for RuR. The reduced incidence of Ca 2+ waves and Ca 2+ sparks in tetracaine reflects the longer period of Ca 2+ uptake required after both events before sufficiently high SR Ca 2+ is reached to activate release. The raised SR Ca 2+ and restoration of the activity of RyR will apply to Ca 2+ release both at the major clusters involved in spark activity and any sub-spark units and this may explain why Ca 2+ waves persist on perfusion of tetracaine. Previous studies using SERCA inhibitors reported a lowered frequency and eventual inhibition of Ca 2+ waves (O Neill et al. 2004) andsparks (Lukyanenkoet al. 2000); the present study reproduced these data and showed that under identical conditions, spark frequency and the proportion of cells exhibiting Ca 2+ waves decreased in parallel. This is consistent with the concept that SR Ca 2+ levels are a major factor in the triggering of both events; reduced SERCA activity will increase the time necessary for SR [Ca 2+ ] to reach a threshold level. Lower SERCA activity has also been reported to reduce the threshold for Ca 2+ release (O Neill et al. 2004), and therefore the net effect of SERCA inhibition will depend on the balance of both these factors. Rogue RyRs One mechanism which could explain the disparity between Ca 2+ wave and Ca 2+ spark inhibition by RuR is the involvement of extra-dyadic or rogue RyRs as an SR Ca 2+ leak pathway as proposed by several groups (Sobie et al. 2006; Lukyanenko et al. 2007; Hayashi et al. 2009). The existence of smaller clusters which do not cause detectable spark events yet may substantially contribute to diastolic leak has been suggested (Sobie et al. 2006; Zima & Blatter, 2009). While not influencing the Ca 2+ spark activity, these smaller clusters may act to facilitate the propagation of the Ca 2+ signal between the large clusters of RyR at the dyad. A simple computational modelofca 2+ wave propagation, based on a previously published description of the cardiac RyR cluster (Sobie et al. 2002), was constructed to examine the relationship between cluster size and the extent of inhibition of SR Ca 2+ release. The simulations showed that inhibition of large (30 RyR) clusters hardly affected spark size, whereas the smaller (5 RyR) clusters were much more sensitive to inhibition. Because depletion of local SR [Ca 2+ ]isfaster and more complete with large RyR clusters, inhibition of a few RyRs in this case has little effect on either the total quantity of Ca 2+ released or the Ca 2+ spark amplitude. This agrees with the relatively small changes in spark size detected in our experiments. This suggests that cardiac cells contain both large and small RyR clusters. Most of the release events identifiable as Ca 2+ sparks result from the coordinated opening of relatively large RyR clusters, and RuR inhibition of these ryanodine receptors can reduce Ca 2+ spark frequency while barely affecting spark amplitude (e.g. at 3 min in Fig. 3C). However, Ca 2+ wave propagation may also require the participation of small clusters of only a few RyRs. Ca 2+ release events produced by these small clusters may be too small to be detected and therefore contribute an invisible SR Ca 2+ leak (Sobie et al. 2006). Their small size, however, makes them more susceptible to moderate degrees of inhibition by RuR, for two reasons. (i) The relationship between total Ca 2+ released and RyR availability (Fig. 5B) is non-linear for large clusters, which implies that with 30 RyRs per cluster, a larger percentage of channels need to be inhibited to reduce released Ca 2+ below a given threshold level. (ii) In addition, simple probabilistic arguments make small RyR clusters more likely to be blocked even when the overall percentage of channels inhibited by RuR is small. This mechanism does not require that rogue RyRs be more sensitive to RuR block than channels in larger clusters. An alternative explanation is that differences in phosphorylation state or regulation by the calsequestrin/triadin/junctin complex may make the rogue RyRs more susceptible to inhibition by RuR. Further work is required to distinguish these possiblities. In summary, this study is the first to demonstrate a dramatically different time course of inhibition of Ca 2+ wave and Ca 2+ spark events in ventricular myocytes. RuR increased SR Ca 2+ levels consistent with a block of SR Ca 2+ leak, but no significant change in Ca 2+ spark size or frequency was observed. Over a similar time period the incidence of Ca 2+ waves was dramatically decreased. These data suggest that another non-spark Ca 2+ leak pathway may be blocked by RuR, and this leak pathway is important in the mechanisms that allow Ca 2+ sparks to initiate and propagate a Ca 2+ wave. The occurrence of Ca 2+ waves during diastole has been linked to the generation of delayed after-depolarisation and subsequent arrhythmias. Our results suggest that there may be a pharmacological strategy that can inhibit the initiation and propagation of these events without dramatically affecting the activity of Ca 2+ release from the larger clusters of RyRs in the dyad that is essential for E C coupling. References Ashley RH & Williams AJ (1990). Divalent cation activation and inhibition of single calcium release channels from sheep cardiac sarcoplasmic reticulum. JGenPhysiol95,

11 J Physiol Ca 2+ wave and Ca 2+ spark events 4741 Bridge JH, Ershler PR & Cannell MB (1999). Properties of Ca 2+ sparks evoked by action potentials in mouse ventricular myocytes. JPhysiol518, Chen-Izu Y, McCulle SL, Ward CW, Soeller C, Allen BM, Rabang C, Cannell MB, Balke CW & Izu LT (2006). Three-dimensional distribution of ryanodine receptor clusters in cardiac myocytes. Biophys J 91,1 13. Cheng H, Lederer WJ & Cannell MB (1993). Calcium sparks: elementary events underlying excitation-contraction coupling in heart muscle. Science 262, Cheng H, Song LS, Shirokova N, Gonzalez A, Lakatta EG, Rios E & Stern MD (1999). Amplitude distribution of calcium sparks in confocal images: theory and studies with an automatic detection method. Biophys J 76, Currie S, Loughrey CM, Craig MA & Smith GL (2004). Calcium/calmodulin-dependent protein kinase IIδ associates with the ryanodine receptor complex and regulateschannelfunctioninrabbitheart.biochem J 377, Drummond GB (2009). Reporting ethical matters in The Journal of Physiology: standards and advice. JPhysiol 587, Franzini-Armstrong C, Protasi F & Ramesh V (1998). Comparative ultrastructure of Ca 2+ release units in skeletal and cardiac muscle. Ann N Y Acad Sci 853, Franzini-Armstrong C, Protasi F & Ramesh V (1999). Shape, size, and distribution of calcium release units and couplons in skeletal and cardiac muscles. Biophys J 77, Gyorke S, Lukyanenko V & Gyorke I (1997). Dual effects of tetracaine on spontaneous calcium release in rat ventricular myocytes. JPhysiol500, Hayashi T, Martone ME, Yu Z, Thor A, Doi M, Holst MJ, Ellisman MH & Hoshijima M (2009). Three-dimensional electron microscopy reveals new details of membrane systems for Ca 2+ signaling in the heart. JCellSci122, Izu LT, Means SA, Shadid JN, Chen-Izu Y & Balke CW (2006). Interplay of ryanodine receptor distribution and calcium dynamics. Biophys J 91, Keizer J & Smith GD (1998). Spark-to-wave transition: saltatory transmission of calcium waves in cardiac myocytes. Biophys Chem 72, Lindsay AR & Williams AJ (1991). Functional characterisation of the ryanodine receptor purified from sheep cardiac muscle sarcoplasmic reticulum. Biochim Biophys Acta 1064, Loughrey CM, MacEachern KE, Neary P & Smith GL (2002). The relationship between intracellular [Ca 2+ ]andca 2+ wave characteristics in permeabilised cardiomyocytes from the rabbit. JPhysiol543, Lukyanenko V, Gyorke I, Subramanian S, Smirnov A, Wiesner TF & Gyorke S (2000). Inhibition of Ca 2+ sparks by ruthenium red in permeabilized rat ventricular myocytes. Biophys J 79, Lukyanenko V, Ziman A, Lukyanenko A, Salnikov V & Lederer WJ (2007). Functional groups of ryanodine receptors in rat ventricular cells. JPhysiol583, MacQuaide N, Dempster J & Smith GL (2007). Measurement and modeling of Ca 2+ waves in isolated rabbit ventricular cardiomyocytes. Biophys J 93, MacQuaide N, Dempster J & Smith GL (2009). Assessment of sarcoplasmic reticulum Ca 2+ depletion during spontaneous Ca 2+ waves in isolated permeabilized rabbit ventricular cardiomyocytes. Biophys J 96, Marx SO, Gaburjakova J, Gaburjakova M, Henrikson C, Ondrias K & Marks AR (2001). Coupled gating between cardiac calcium release channels (ryanodine receptors). Circ Res 88, McIntosh MA, Cobbe SM & Smith GL (2000). Heterogeneous changes in action potential and intracellular Ca 2+ in left ventricular myocyte sub-types from rabbits with heart failure. Cardiovasc Res 45, O Neill SC, Miller L, Hinch R & Eisner DA (2004). Interplay between SERCA and sarcolemmal Ca 2+ efflux pathways controls spontaneous release of Ca 2+ from the sarcoplasmic reticulum in rat ventricular myocytes. JPhysiol559, Rousseau E & Meissner G (1989). Single cardiac sarcoplasmic reticulum Ca 2+ -release channel: activation by caffeine. Am J Physiol Heart Circ Physiol 256, H328 H333. Santiago DJ, Curran JW, Bers DM, Lederer WJ, Stern MD, Rios E & Shannon TR (2010). Ca sparks do not explain all ryanodine receptor-mediated SR Ca leak in mouse ventricular myocytes. Biophys J 98, Smith GD, Keizer JE, Stern MD, Lederer WJ & Cheng H (1998). A simple numerical model of calcium spark formation and detection in cardiac myocytes. Biophys J 75, Smith GL & O Neill SC (2001). A comparison of the effects of ATP and tetracaine on spontaneous Ca 2+ release from rat permeabilised cardiac myocytes. JPhysiol534, Sobie EA, Dilly KW, dos Santos CJ, Lederer WJ & Jafri MS (2002). Termination of cardiac Ca 2+ sparks: an investigative mathematical model of calcium-induced calcium release. Biophys J 83, Sobie EA, Guatimosim S, Gomez-Viquez L, Song LS, Hartmann H, Saleet JM & Lederer WJ (2006). The Ca 2+ leak paradox and rogue ryanodine receptors: SR Ca 2+ efflux theory and practice. Prog Biophys Mol Biol 90, Sobie EA, Song LS & Lederer WJ (2005). Local recovery of Ca 2+ release in rat ventricular myocytes. JPhysiol565, Soeller C, Crossman D, Gilbert R & Cannell MB (2007). Analysis of ryanodine receptor clusters in rat and human cardiac myocytes. Proc Natl Acad Sci U S A 104, Trafford AW, Sibbring GC, Diaz ME & Eisner DA (2000). The effects of low concentrations of caffeine on spontaneous Ca release in isolated rat ventricular myocytes. Cell Calcium 28, Weir WG & Hess P (1984). Excitation-contraction coupling in cardiac purkinje fibers: Effects of cardiotonic steroids on the intracellualr [Ca 2+ ] transient, membrane potential, and contraction. JGenPhysiol83,

12 4742 N. MacQuaide and others J Physiol Xu L, Tripathy A, Pasek DA & Meissner G (1998). Potential for pharmacology of ryanodine receptor/calcium release channels. Ann N Y Acad Sci 853, Yin CC, Blayney LM & Lai FA (2005). Physical coupling between ryanodine receptor-calcium release channels. JMol Biol 349, Zima AV & Blatter LA (2009). Properties of sarcoplasmic reticulum Ca leak in rabbit ventricular and atrial myocytes. Biophys J 96, 276a 277a. Author contributions N.M., experimental work, data analysis and writing. H.E.R. computational modelling. E.S. computational modelling and writing. G.L.S. experimental design and writing. Acknowledgements The authors acknowledge the financial support of the British Heart Foundation (RG/04/07) to GLS and the National Institutes of Health (HL076230) to EAS.